Good Practice in CNN Feature Transfer

By Liang Zheng, Yali Zhao, Shengjin Wang, Jingdong Wang, Qi Tian. 2016

175 citations recorded by the source corpus.

The objective of this paper is the effective transfer of the Convolutional Neural Network (CNN) feature in image search and classification. Systematically, we study three facts in CNN transfer. 1) We demonstrate the advantage of using images with a properly large size as input to CNN instead of the conventionally resized one. 2) We benchmark the performance of different CNN layers improved by average/max pooling on the feature maps. Our observation suggests that the Conv5 feature yields very competitive accuracy under such pooling step. 3) We find that the simple combination of pooled features extracted across various CNN layers is effective in collecting evidences from both low and high level descriptors. Following these good practices, we are capable of improving the state of the art on a number of benchmarks to a large margin.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yi Yang
2. Martial Hebert
3. Qi Tian
4. Yu-Xiong Wang
5. Liang Zheng
6. Deva Ramanan
7. Trevor Darrell
8. Luc Van Gool

In Computer Vision and Pattern Recognition

1. Yi Yang
2. Luc Van Gool
3. Qi Tian
4. Dacheng Tao
5. Liang Zheng
6. Xiaogang Wang
7. Trevor Darrell
8. Ming-Hsuan Yang

In Representation & Self-Supervised Learning

1. Martial Hebert
2. Deva Ramanan
3. Yu-Xiong Wang
4. Yi Yang
5. Liang Zheng
6. Giorgos Tolias
7. Ondřej Chum
8. Filip Radenovic

Citations in the corpus

• Growing a Brain: Fine-Tuning by Increasing Model Capacity (2017, background): “While there is a large body of work on transfer learning, much of it assumes a fixed capacity model [32, 3, 6, 58, 15].”
• CNN Image Retrieval Learns from BoW: Unsupervised Fine-Tuning with Hard Examples (2016, methodology): “A variety of previous methods apply CNN activations on the task of image retrieval [22,15,23,24,25,43].”
• Fine-Tuning CNN Image Retrieval with No Human Annotation (2017, background): “the task of image retrieval [8], [9], [10], [11], [12], [45].”
• Selective Convolutional Descriptor Aggregation for Fine-Grained Image Retrieval (2016, background): “Very recently, the authors of [26] investigated several effective usages of CNN activa-”
• Speeding up the Hyperparameter Optimization of Deep Convolutional Neural Networks (2018, methodology): “6,7 If hyperparameter values for the same images in di®erent resolutions are similar to each other, we can use this to ¯nd appropriate hyperparameters on images with low resolution and then ¯ne-tune them for the same images with high resolution.”
• SIFT Meets CNN: A Decade Survey of Instance Retrieval (2016, methodology): “t the corresponding filter [9], [10], [134].”